let sutheesh = Engineer(
focus: "Mobile-first engineering × Agentic AI",
platforms: [.iOS, .android],
currentWork: [
"SwiftMCP — bridging Apple Foundation Models to MCP servers",
"OpenChat V2 — agentic AI with AG-UI, A2UI, and streaming",
"RuleGen — auto-generated AI coding rules from codebases"
],
core: [
"On-device ML (Core ML, TensorFlow Lite)",
"Protocol-level systems (MCP, AG-UI, A2UI)",
"Payments & fintech mobile infrastructure"
],
philosophy: "Understand the protocol. Build from primitives. Ship it."
)| Platform | Stack |
|---|---|
| iOS | Swift, SwiftUI, UIKit, Core ML, Combine, Swift Concurrency |
| Android | Kotlin, Jetpack Compose, Room, Coroutines, TensorFlow Lite |
| Cross-Platform | React Native, on-device inference, biometric auth |
| Distribution | App Store Connect, Google Play Console, CI/CD pipelines |
Building at the intersection of mobile engineering and agentic AI — focused on protocols over abstractions.
| Area | Details |
|---|---|
| MCP | Built SwiftMCP — connect Apple Foundation Models to any MCP server in 3 lines of Swift |
| AG-UI / A2UI | OpenChat V2 — real-time agentic streaming with FastAPI + Groq (Llama 4 Scout) + React |
| On-Device ML | Core ML model integration, TensorFlow Lite for Android, edge inference pipelines |
| RAG & Embeddings | ChromaDB, vector search, semantic memory for conversational agents |
| AI Coding Tools | RuleGen — auto-generate coding rules from your codebase for AI agents |
🔗 SwiftMCP
Bridge Apple Foundation Models → MCP servers in 3 lines of Swift
On-device Apple Intelligence meets MCP. Any tool, any server — native Swift integration with LanguageModelSession.
Agentic AI chat with real-time streaming and dynamic UI generation
FastAPI backend + Groq Llama 4 Scout + React frontend. AG-UI implemented as raw NDJSON events — no SDK abstraction layer.
🛡️ RuleGen
Auto-generate AI coding rules from your codebase
Like SwiftLint for AI agents — install once, rules stay in sync with your codebase forever.
Languages
Mobile
Backend & Infrastructure
AI / ML
Exploring production-grade agentic AI infrastructure — bridging the gap between research and real-world deployment:
- 🛡️ MCP Security — trust frameworks for tool-calling agents in regulated environments
- 👁️ Agent Observability — runtime monitoring and tracing for multi-step agent workflows
- 🔄 Self-Healing Agents — autonomous detection and repair of broken agent pipelines
- 📱 Mobile MCP — adapting the Model Context Protocol for resource-constrained mobile devices
- 🧠 Memory Poisoning Defense — protecting persistent agent memory from adversarial manipulation
- 🔗 Building SwiftMCP — connecting Apple's on-device models to the MCP ecosystem
- 🤖 Shipping agentic AI systems with AG-UI streaming and protocol-first architecture
- 📱 Advancing on-device ML for iOS and Android
- 🔬 Researching MCP security, agent observability, and self-healing agent workflows
- ✍️ Writing about AI systems — from foundational concepts to production deployment
- 💬 Ask me about Swift, Kotlin, MCP, on-device ML, agentic AI, mobile architecture
- 🎓 IEEE Senior Member
- 🔬 Peer reviewer for AI/ML conference submissions
- ✍️ Author — Understanding Artificial Intelligence (technical book, 20 chapters)
Engineer driven by curiosity — continuously learning, building, and adding meaningful ideas to the canvas of technology.